scientific article; zbMATH DE number 4001882

From MaRDI portal
Publication:3754444





[For the entire collection see Zbl 0595.00027.] The aim of this paper is to analyze computational complexity of a class of methods for solving semiinfinite programming problems: \[ (1)\quad \min_{x\in X}f(x),\quad X=\{x\in R^ n| \quad c(x,t)\leq 0,\quad t\in C\} \] where \(f: R^ n\to R\), \(c: R^ n\times C\to R\) and \(C\subseteq R^ r\) is a compact index set. Problem (1) is a hard problem and the majority of the existing methods are either based on discretization (which involves consideration of exponentially growing number of grid points) or require global maximization problems (over the index set C) to be solved in order to perform the feasibility check. There are also methods which combine these two approaches. We have developed a class of methods based on 'selective' discretization. Considerable numerical evidence indicates that these methods effectively use only a linearly growing number of grid points. In this paper we give theoretical grounds for such a behaviour.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3754444)